AI-related debt has been rising rapidly in corporate bonds, private credit and project finance and is on track to exceed $800 billion this year, Federal Reserve Vice Chair for Supervision Michelle Bowman told the Council on Foreign Relations on Tuesday. Other obligations, in the form of long-term lease and contract commitments and residual value guarantees, have also surged and their potential payments are estimated at more than three times the amount of debt, she said, with highly rated hyperscalers assuming off-balance-sheet liabilities that let smaller and riskier companies use them as collateral in asset-based financing for data centers and computing equipment, and hyperscalers and chip designers making equity investments in smaller companies that buy their products, creating the appearance of circular financing.
AI revenues are rising quickly but are not yet sufficient to pay for maintaining existing models and data centers and developing new ones, Bowman said. The largest US bank holding companies have some exposure to AI-related financing, scrutinized in the annual supervisory stress tests, and life insurers, pension funds, private credit and mutual funds are exposed as well, but the complex structures make counterparty risk difficult to assess properly. If AI-related revenues fall short of sharply rising costs, she warned, the result could be a sharp correction in AI equity prices, higher financing costs and reduced access to credit, slowing the buildout itself.